RobustPCBWhere Robust Design Starts
CalculatorsالعربيةUpgradeSign inSign up

What's new

Full changelogEngine 1.26.4

RC Snubber for Switch-Node Ringing

Power & Regulators

Which R and C stop this ringing, and can the resistor take the heat?

Inputs

Read on the scope, with the probe tip and a ground spring at the switch.
Hz
150 MHz
A C0G part right across the switch. Pick it to roughly halve the ring (about 3 × Cpar, Kollman).
F
1 nF
Hz
75 MHz
For a flyback, Vin plus the reflected output (Ridley). Use the highest it reaches.
V
Hz
500 kHz
The converter's rated output.
W
Kollman starts at 3. Ridley starts at 1 (XC = R at the ring). More C damps harder and costs more power.
Typical thick-film rating at 70 °C. Check your part.
Derating
From the derating curve at your ambient, and your own policy.
%

Results

Snubber R+Cs
3.3Ω + 1.0nF
Pass

Checked: PR ≤ 0.8 × Prated × d (design margin): 288 mW is 0.576 × the 500 mW you allow, 1 W × 50 %
Show working
  1. Parasitic capacitanceC_par = C_add / ((f_1 / f_2)² − 1) = 1 nF / ((150 MHz / 75 MHz)² − 1) = 333.3 pF
  2. Parasitic inductanceL_par = 1 / ((2π × f_1)² × C_par) = 1 / ((2π × 150 MHz)² × 333.3 pF) = 3.377 nH
  3. Damping resistorZ_0 = √(L_par / C_par) = √(3.377 nH / 333.3 pF) = 3.183 Ω, nearest E24 3.3 Ω
  4. Snubber capacitorC_s = k × C_par = 3 × 333.3 pF = 1 nF, rounded up in E12 to 1.0 nF
  5. Resistor powerP_R = C_s × V_sw² × f_sw = 1.0 nF × 24² × 500 kHz = 288 mW
Snubber power PR
288mW
Efficiency cost PR/Pout
1.2%
Characteristic impedance Z0
3.183Ω
All results (7)
Parasitic capacitance Cpar
333.3pF
Parasitic inductance Lpar
3.377nH
Ring frequency f1
150MHz
R and Cs before rounding
3.183Ω + 1nF
Ring over switching frequency f1/fsw
300:1
Smallest passing package
Overshoot estimate 

lumped model, estimate. The same model with no snubber rings to 100 %.
43.55%
With the snubber, lumped-model estimateNo snubber, lossless estimate

Accuracy

Source
  • R. Kollman
  • R. Ridley

R. Kollman, Power Tip #16: Snubbing the forward converter, EDN the equation

R. Ridley, Flyback Snubber Design, Switching Power Magazine the equation

Precision
Kollman's algebra is exact for a fixed L and a linear C, so C_par and L_par are as good as the two ring readings, or as the L and C you enter.

R = Z_0 and C_s = k·C_par are the published starting points, not an optimum. P_R is Ridley's conservative estimate, and the overshoot comes from a lumped model of one L and one C, so it is an estimate.

Verified against

2 independent cases. See the working.

This is a design aid. The engineer remains responsible for the design and for checking the standard itself.

Parameter sweep

Vary one input over a range and see the answer and verdict at each step, as a table and a curve.Pro

Worst-case corners

Put a tolerance on each input and get the worst-case band around the answer.Pro

Circuit

The principle

The loop inductance and the capacitance at the switch node form a resonant tank. Every switching edge rings it at f1, which overshoots the switch and radiates. An R–C across the switch damps the ring: R close to the tank's characteristic impedance absorbs it, and Cs in series keeps the switching-frequency voltage off R (Ridley). This page finds the parasitics from two ring readings or from known L and C, picks R and Cs, and checks the resistor's power.

Parasitic C from the added capacitor
Cpar=Cadd(f1/f2)2−1C_{par} = \frac{C_{add}}{(f_1/f_2)^{2} - 1}
Kollman's method: read the ring at f1, then again at f2 with Cadd across the switch. Exact for a fixed L and a linear C.
Parasitic L
Lpar=1(2πf1)2 CparL_{par} = \frac{1}{(2\pi f_1)^{2}\, C_{par}}
With known L and C the page works the other way, f1 = 1/(2π√(Lpar·Cpar)).
Damping resistor R = Z_0
R=Z0=LparCparR = Z_0 = \sqrt{\frac{L_{par}}{C_{par}}}
Ridley. The page takes the nearest E24 value.
Snubber capacitor C_s = k·C_par
Cs=k CparC_s = k\, C_{par}
k = 1 is Ridley's XC = R at f1, since 1/(2π·f1·Z0) = Cpar. k = 3 is Kollman's starting point. The page rounds Cs up in E12.
Resistor power P_R = C_s V² f_sw
PR=Cs Vsw2 fswP_R = C_s\, V_{sw}^{2}\, f_{sw}
Ridley. No ½: R takes the energy on the charge and again on the discharge. Ridley calls it a conservative estimate.
Efficiency cost
PRPout\frac{P_R}{P_{out}}
Pass up to 1 %, fail above 2 %: design margins, not a standard.
  • f1f_1ring frequency with no snubber
  • f2f_2ring frequency with Cadd across the switch
  • CaddC_{add}the test capacitor added across the switch
  • CparC_{par}switch-node capacitance: Coss, the diode and stray
  • LparL_{par}loop inductance, or the leakage inductance
  • Z0Z_0characteristic impedance of the tank
  • kksnubber capacitance as a multiple of Cpar
  • VswV_{sw}voltage across the switch when off
  • R, CsR,\,C_sthe snubber resistor and capacitor
  • fswf_{sw}switching frequency
  • PRP_Rpower in the snubber resistor
  • PoutP_{out}converter output power
More detail

Reading the parasitics

Read f1 with the probe tip and a short ground spring right at the switch, so the probe adds as little as it can. Then put a C0G capacitor across the switch and read f2. Kollman picks one that about halves the ring, which is 3 × Cpar. The two readings give Cpar and Lpar.

Choosing R and Cs

R = Z0 damps the tank (Ridley). Ridley sizes Cs so its reactance at f1 equals R, which is k = 1; Kollman starts at 3 × Cpar. More C damps harder and costs more power, since PR grows with Cs. Keep the ring at least two orders above fsw, or the loss grows (Ridley).

The resistor

The package ratings in the list are typical thick-film figures at 70 °C, not a standard. Check your part's datasheet, or type its rating as custom, and take the allowed share d from its derating curve. PR passes up to 0.8 × the rating × d and fails above the rating × d, a design margin.

This checks the average power. At each turn-on Cs, charged to Vsw, also discharges through R in a short pulse of up to Vsw/R, so check the part's pulse rating too.

The overshoot estimate

The chart and the overshoot come from a lumped model: a step of Vsw drives Lpar into Cpar in parallel with R and Cs in series, from rest, stepped in time (fourth-order Runge–Kutta, 1 000 steps a ring period, five periods). With no snubber the same model rings to twice Vsw. It leaves out the switch's own edge, Coss changing with voltage and the losses in the loop, so confirm the result on the scope.

Where the model stops

With two ring readings, Cadd must be at least Cpar (f1/f2 of √2 or more), or the page marks the answer Unverified: below that a 1 % error in f1/f2 moves Cpar by more than 4 %. The threshold is this page's choice. The answer is Unverified too when the readings give a Cpar outside 1 pF to 100 nF or an Lpar outside 10 pH to 100 µH, the range the Known L and C fields accept. With known L and C nothing is flagged, because R and Cs follow directly from what you enter; PR is still Ridley's conservative estimate, since the inductance does some of the charging and the true loss can be a little lower.

Measure the ring at the switch, start from R = Z0 and Cs = 3 × Cpar, and rate the resistor for CsVsw²fsw.
References: R. Ridley, Flyback Snubber Design, Switching Power Magazine. R. Kollman, Power Tip #16: Snubbing the forward converter, EDN: the add-a-capacitor method and k ≈ 3. Related: Buck Converter Design, MOSFET Loss Budget, Input EMI Filter & Damping.

Engine version ⁨1.26.4⁩